High-level workflow:

• Create a Cognitive Search service in Azure.

• Index your database (Orders table with Comments). You can connect directly to SQL DB or push documents programmatically.

• Query the index from your C# app using the Azure.Search.Documents SDK.

• Present results (matching orders or summaries).

This example assumes you have:

• A Cognitive Search service

• An index called orders

• A field comments inside that index

using System;
using Azure;
using Azure.Search.Documents;
using Azure.Search.Documents.Models;class Program
{
    static void Main(string[] args)
    {
        string serviceEndpoint = "https://<your-search-service>.search.windows.net";
        string indexName = "orders";
        string apiKey = "<your-admin-or-query-key>";       

             var client = new SearchClient(
            new Uri(serviceEndpoint),
            indexName,
            new AzureKeyCredential(apiKey));       

             Console.WriteLine("Enter your question:");
        string userInput = Console.ReadLine();       

             // Step 1: Run a semantic search query
        var options = new SearchOptions()
        {
            QueryType = SearchQueryType.Semantic,
            Size = 10, // limit results
            QueryLanguage = "en-us"
        };       

             // Optional: select specific fields
        options.Select.Add("orderId");
        options.Select.Add("comments");       

             // Step 2: Execute the search
        SearchResults<SearchDocument> response = client.Search<SearchDocument>(userInput, options);                   

             // Step 3: Display results
        Console.WriteLine("Results:");
        foreach (SearchResult<SearchDocument> result in response.GetResults())
        {
            Console.WriteLine($"OrderId: {result.Document["orderId"]}");
            Console.WriteLine($"Comments: {result.Document["comments"]}");
            Console.WriteLine($"Score: {result.Score}\n");
        }
    }
}

Example of queries a user could have:

• “Show me all orders where delivery is delayed due to an invalid address”

• “List comments mentioning weather problems”

• “How many orders had issues with customer service?”

Cognitive Search retrieves the matching documents, and you can do post-processing (like counting, grouping) in your app.

Key Difference vs. OpenAI + SQL

• OpenAI + SQL: More flexible, can aggregate and summarize directly via SQL translation.

• Cognitive Search: Strong at semantic/document retrieval, but summarization/aggregation may require another step (e.g., call Azure OpenAI to summarize the retrieved results).

We could use a combination of these two approaches! This is also called RAG - Retrieval Augmented Generation.

Comparison